Question

I am using pandas to get the count of the Text type data and to find out the top 5 among the given data.

Input file is as follows:

Gears of war 3
Gears of war
Assassin creed
.......
.......
Crysis 2
Gears of war3
Sims

My Output is as follows:

{
    'Gears of War 3': 6,
    'Batman': 5,
    'gears of war 3': 4,
    'Rocksmith': 5,
    'nan': 32870
}

I want my code to skip counting nan values in my csv file.

My code is as follows:

data = pandas.read_csv('D:\my_file.csv')

for colname, dtype in data.dtypes.to_dict().iteritems():
    if dtype == 'object':
        print colname
        count = Counter(data[colname])
        d = dict((str(k), v) for k, v in count.iteritems())
        f = dict(sorted(d.iteritems(), key=lambda item: item[1], reverse = True)[:5])

Answer 1

use value_counts() to count the non-Nan values:

        one       two     three
a  0.196508 -0.465768 -0.710062
b       NaN       NaN       NaN
c  0.532602  1.835499  0.465047
d       NaN       NaN       NaN
e  0.175336 -0.471934 -1.517812
f -2.392756 -0.021330 -0.239647
g       NaN       NaN       NaN
h -0.612554  0.238613 -1.060184

df2['one'].value_counts()

 0.532602    1
 0.196508    1
 0.175336    1
-0.612554    1
-2.392756    1